Association Between Expiratory Central Airway Collapse and Respiratory Outcomes Among Smokers

Importance  Central airway collapse greater than 50% of luminal area during exhalation (expiratory central airway collapse [ECAC]) is associated with cigarette smoking and chronic obstructive pulmonary disease (COPD). However, its prevalence and clinical significance are unknown.

Objective  To determine whether ECAC is associated with respiratory morbidity in smokers independent of underlying lung disease.

Design, Setting, and Participants  Analysis of paired inspiratory-expiratory computed tomography images from a large multicenter study (COPDGene) of current and former smokers from 21 clinical centers across the United States. Participants were enrolled from January 2008 to June 2011 and followed up longitudinally until October 2014. Images were initially screened using a quantitative method to detect at least a 30% reduction in minor axis tracheal diameter from inspiration to end-expiration. From this sample of screen-positive scans, cross-sectional area of the trachea was measured manually at 3 predetermined levels (aortic arch, carina, and bronchus intermedius) to confirm ECAC (>50% reduction in cross-sectional area).

Exposures  Expiratory central airway collapse.

Main Outcomes and Measures  The primary outcome was baseline respiratory quality of life (St George’s Respiratory Questionnaire [SGRQ] scale 0 to 100; 100 represents worst health status; minimum clinically important difference [MCID], 4 units). Secondary outcomes were baseline measures of dyspnea (modified Medical Research Council [mMRC] scale 0 to 4; 4 represents worse dyspnea; MCID, 0.7 units), baseline 6-minute walk distance (MCID, 30 m), and exacerbation frequency (events per 100 person-years) on longitudinal follow-up.

Results  The study included 8820 participants with and without COPD (mean age, 59.7 [SD, 6.9] years; 4667 [56.7%] men; 4559 [51.7%] active smokers). The prevalence of ECAC was 5% (443 cases). Patients with ECAC compared with those without ECAC had worse SGRQ scores (30.9 vs 26.5 units; P < .001; absolute difference, 4.4 [95% CI, 2.2-6.6]) and mMRC scale scores (median, 2 [interquartile range [IQR], 0-3]) vs 1 [IQR, 0-3]; P < .001]), but no significant difference in 6-minute walk distance (399 vs 417 m; absolute difference, 18 m [95% CI, 6-30]; P = .30), after adjustment for age, sex, race, body mass index, forced expiratory volume in the first second, pack-years of smoking, and emphysema. On follow-up (median, 4.3 [IQR, 3.2-4.9] years), participants with ECAC had increased frequency of total exacerbations (58 vs 35 events per 100 person-years; incidence rate ratio [IRR], 1.49 [95% CI, 1.29-1.72]; P < .001) and severe exacerbations requiring hospitalization (17 vs 10 events per 100 person-years; IRR, 1.83 [95% CI, 1.51-2.21]; P < .001).

Conclusions and Relevance  In a cross-sectional analysis of current and former smokers, the presence of ECAC was associated with worse respiratory quality of life. Further studies are needed to assess long-term associations with clinical outcomes.

Expiratory central airway collapse (ECAC) is defined by excessive airway collapse during expiration, resulting from either cartilaginous weakening or redundancy of the posterior membranous wall.¹ The frequency of ECAC, which includes tracheobronchomalacia and excessive dynamic airway collapse, has not been assessed in large studies; using bronchoscopy for ECAC diagnosis, the prevalence in people with known respiratory problems has ranged from 1% to 53%.^2-4

With increasing use of noninvasive imaging techniques such as computed tomography (CT), ECAC is increasingly recognized in the adult population, especially in association with cigarette smoking and chronic obstructive pulmonary disease (COPD).⁴ While the small conducting airways less than 2 mm in diameter are the major site of resistance to airflow in COPD,⁵ collapse greater than 50% of the larger central airway luminal area during exhalation has been hypothesized to cause additional airflow obstruction and respiratory morbidity.^3,6-8 Although ECAC in adults has been associated with cough and dyspnea in previous smaller studies, it is not known if this is attributable to underlying disease processes such as chronic bronchitis and emphysema or to ECAC.^3,6,7 In addition, data about the prevalence of ECAC and correlation with overall respiratory quality of life and other clinical outcomes are limited.

We analyzed paired inspiratory-expiratory CT images of participants in a cohort of current and former smokers enrolled in the COPDGene study to determine the prevalence and clinical significance of ECAC. We hypothesized that ECAC was common in patients with COPD and in smokers without airflow obstruction and was associated with respiratory morbidity independent of underlying lung disease.

Current and former smokers aged 45 to 80 years from a large multicenter cohort study (COPDGene) were included in the study. Details of COPDGene have been published.⁹ The COPDGene study was approved by the institutional review boards of all 21 participating centers, and all participants provided written informed consent prior to participation.

Participants were enrolled from local communities across the United States. Those with known lung disease other than COPD and asthma were excluded. Demographic characteristics such as age and sex were collected at enrollment. We collected information on race based on participants’ self-identification as non-Hispanic white or non-Hispanic African American race. Participants from these races were specifically recruited to meet the requirements for the planned genome-wide association study analyses of genetic susceptibility to cigarette smoke. Spirometry was performed before and after bronchodilator administration, in accordance with American Thoracic Society criteria.¹⁰ Computed tomography scans were performed at maximal inspiration (total lung capacity), end-tidal expiration (functional residual capacity), and, at 1 center, at residual volume. Emphysema (percentage of lung volume at total lung capacity with attenuation less than −950 Hounsfield units [HU] [%LAA950_insp, where LAA indicates low attenuation area]) and gas trapping (percentage of lung volume at functional residual capacity [or residual volume] with attenuation less than −856 HU [%LAA856_exp]) were quantitated using 3D Slicer software (http://airwayinspector.acil-bwh.org), and airway dimensions were measured using Pulmonary Workstation 2 (VIDA Diagnostics).⁹ Wall area percentage of segmental airways was used to quantify airway disease.⁹ One center acquired expiratory CT scans at residual volume, and these cases were analyzed separately (eTable 1 and eTable 2 in the Supplement).

The primary outcome was respiratory disease–related health impairment and quality of life, assessed using St George’s Respiratory Questionnaire (SGRQ) scores.¹¹ The SGRQ ranges from 0 to 100 (with 100 indicating the worst quality of life); the minimum clinically important difference (MCID) is 4 units. Secondary outcomes were dyspnea, exercise capacity, and exacerbation frequency. Dyspnea was measured using the modified Medical Research Council (mMRC) dyspnea score.¹² mMRC scores range from 0 for minimal symptoms to 4 for severe dyspnea; there is no accepted MCID. However, based on the distribution of this score in our cohort, we defined the MCID for mMRC as 0.5 SD [0.7 units].¹³ Six-minute walk distance was assessed according to American Thoracic Society guidelines, with an MCID of 30 m.^9,14

Follow-up data were obtained by contacting participants every 3 to 6 months through an automated telecommunication system using a validated questionnaire.¹⁵ Exacerbations were defined using a modified version of the Epidemiology Standardization Project questionnaire (American Thoracic Society–Division of Lung Diseases 78).^16,17 We defined exacerbations as respiratory worsening (increase in dyspnea, cough, or sputum production) lasting at least 48 hours and requiring use of either antibiotics or systemic steroids, as reported by the participants during follow-up telephone encounters. Exacerbations were classified as “severe” when they resulted in hospitalization. All-cause mortality was also compared by ECAC status at enrollment. Study participants and coordinators were blinded to the presence or absence of ECAC.

We performed CT analysis of ECAC in 2 stages (eFigure 1 in the Supplement). Measurements were performed on baseline CT scans by investigators blinded to the participants' clinical characteristics. Participants with paired inspiratory-expiratory scans that passed quality control were included. In the first stage, a quantitative method using Pulmonary Workstation 2 was used to screen paired inspiratory-expiratory scans and assess percent change in minor axis diameter of the trachea. Measurements were made for the inner diameter of the trachea, including major and minor axis diameters, at every 10 mm from the carina caudocranially. Participants with at least 30% reduction in minor axis diameter from inspiration to expiration were categorized as screen-positive for ECAC, assuming that a 30% reduction in diameter corresponds geometrically to a 50% reduction in cross-sectional area (CSA). In addition, to include more patients, participants diagnosed as having tracheobronchomalacia on visual reading, either by local radiologists at each site of enrollment or during a CT workshop, were included.¹⁸ This workshop involved 58 radiologists and pulmonologists evaluating the chest CT scans to describe abnormalities as well as to evaluate concordance between visual and quantitative measurements.

In the second stage, screen-positive scans were evaluated for confirmation of ECAC by 3 readers (2 chest radiologists and 1 pulmonologist). Participants with negative scans on screening were deemed to be controls without ECAC. Window levels were set at −550 to −700 HU, and window width at 1200-1500 HU. CSA was measured manually using standard DICOM software at 3 levels in both inspiratory and expiratory scans: at the level of the aortic arch (at the origin of the subclavian artery), at the level of the carina, and at the level of the bronchus intermedius, just distal to the origin of the right upper lobe bronchus (Figure). Percent airway collapse was assessed using the formula [(CSA at end-inspiration − CSA at end-expiration)/CSA at end-inspiration]. ECAC was defined as 50% or greater reduction in CSA at any level based on the final manual reading. We selected a 50% threshold for defining ECAC based on previous studies as well as our data showing that differences in SGRQ meet the MCID of 4 units at and beyond this threshold (eFigure 2 in the Supplement). All readers scanned the entire length of the trachea to assess qualitatively if there was a greater than 50% reduction in CSA at any other level. Scans without a 50% reduction in CSA at any level on manual measurement were deemed controls. A subset of 300 overlapping scans (n = 100 each) were read by each reader to calculate interobserver and intraobserver variability. The original reader’s measurement of airway collapse was used in the analysis.

Baseline data are expressed as means with standard deviations for normally distributed values. Intraobserver and interobserver agreements for diagnosis of ECAC were calculated using Cohen κ. The intraclass correlation coefficient was calculated to assess interrater reliability. Bivariable comparisons were made between those with and without ECAC using χ² test for categorical variables and 2-tailed independent t tests for continuous variables. Variables significant on bivariable analyses at a 2-sided α of .05 were included in a multivariable model, and logistic regression was performed to identify variables independently associated with ECAC. Multivariable linear regression analyses were performed to assess relationships between ECAC and respiratory morbidity indices such as mMRC, SGRQ, and 6-minute walk distance, using age, sex, race, body mass index (BMI), forced expiratory volume in the first second (FEV₁), pack-years of smoking, emphysema, gas trapping, and segmental wall area percentage as covariates.

To assess the differences in exacerbations on follow-up, negative binomial regression models allowing for multiple events in the same participant were created with adjustment for age, race, sex, BMI, smoking burden, FEV₁, and emphysema. To assess the robustness of the negative binomial models, we also compared the time to first contact at which an exacerbation was captured by the longitudinal follow-up system using Cox proportional hazards models. The time-to-event analyses were adjusted for age, race, sex, BMI, pack-years of smoking, FEV₁, and emphysema on CT. Cox proportional hazards analyses were also performed to compare mortality by ECAC status after adjustment for age, race, sex, BMI, smoking pack-years, FEV₁, emphysema on CT, gas trapping on CT, and segmental wall area percentage. Participants who agreed to be part of the long-term follow-up cohort were analyzed for exacerbations and mortality. Patients were censored at the time they last reported outcomes in the long-term follow-up system. We controlled for multiple comparisons within the entire cohort as well as the COPD and non-COPD subgroups using the false discovery rate procedure.¹⁹

All analyses were performed using SSPS version 22.0 (SPSS Inc). P<.05 (2-sided) was considered statistically significant.

The study included 8820 participants with a mean age of 59.7 (SD, 9.0) years (eFigure 1 in the Supplement). Of these, 4667 (56.7%) were men, 5428 (66.0%) were white, 2792 (34.0%) were African American, 4559 (51.7%) were active smokers, and 3856 (43.7%) had COPD diagnosed by ratio of FEV₁ to forced vital capacity less than 0.70. The κ values for intraobserver and interobserver agreement for detecting ECAC were 0.77 (95% CI, 0.71 to 0.83) and 0.73 (95% CI, 0.67 to 0.79), respectively. The intraclass correlation coefficient between raters was 0.95 (95% CI, 0.92 to 0.96) at the aortic arch, 0.91 (95% CI, 0.88 to 0.93) at the carina, and 0.92 (95% CI, 0.88 to 0.94) at the bronchus intermedius.

Four hundred forty-three cases of ECAC were identified (prevalence, 5.0%). The prevalence was higher in participants with COPD than in those without COPD (5.9% [229/3856] vs 4.3% [205/4964]; absolute difference, 1.6% [95% CI, 0.9% to 2.7%]; P = .001), and increased with higher GOLD (Global Initiative for Chronic Obstructive Lung Disease) stage²⁰ (4.8% [33/683] for stage 1, 5.6% [93/1666] for stage 2, 6.6% [66/995] for stage 3, and 7.2% [37/512] for stage 4; P < .001). Compared with controls, participants with ECAC were older (mean, 65 [SD, 8.6] vs 59.4 (SD, 9] years; absolute difference, 5.6 [95% CI, 4.8 to 6.4]; P < .001). The prevalence of ECAC was greater in women than in men (7.2% [297/4153] vs 3.1% [146/4667]; absolute difference, 4.2% [95% CI, 3.1% to 5.0%]; P < .001) and in white compared with African American participants (6.2% [374/6028] vs 2.5% [69/2792]; P < .001; absolute difference, 3.7% [95% CI, 2.9% to 4.6%]; P < .001). Table 1 reports baseline demographics for participants with and without ECAC. Those with ECAC had a higher BMI, higher prevalence of chronic bronchitis, and a greater pack-years of smoking. Participants with ECAC also had greater baseline airflow obstruction (Table 1).

Participants with ECAC had greater percentage of emphysema on CT compared with controls. However, there were no differences in gas trapping and airway wall thickness of either segmental or subsegmental airways (Table 1). Whereas on bivariable analysis, emphysema, and smoking burden were associated with ECAC, on multivariable analyses, ECAC was independently associated with older age, female sex, white race, higher BMI, and greater airflow obstruction (Table 2).

Compared with controls, participants with ECAC had worse respiratory quality of life (SGRQ score, 30.9 [SD, 22.4] vs 26.5 [22.8]; P < .001) at enrollment, an absolute difference of 4.4 units (95% CI, 2.2 to 6.6). After multivariable adjustment, ECAC was associated with worse respiratory quality of life (Table 3).

Those with ECAC had greater dyspnea and worse functional capacity, as measured by lower 6-minute walk distance (Table 1). After multivariable adjustment, ECAC was associated with greater dyspnea but not with lower 6-minute walk distance (Table 3).

We had follow-up data on 7456 participants (413/443 with ECAC [93%] and 7043/8377 without ECAC [84%]) (median follow-up, 4.3 years [range, 0.2 to 6.7 years]). Overall, after multivariable adjustment, participants with ECAC had increased frequency of total (58 vs 35 events per 100 person-years; incidence rate ratio [IRR], 1.49 [95% CI, 1.29 to 1.72]; P < .001) and severe exacerbations requiring hospitalization (17 vs 10 events per 100 person-years; IRR, 1.83 [95% CI, 1.51 to 2.21]; P < .001).

Among participants with COPD (n=3388; median follow-up, 4.3 years [interquartile range {IQR}, 3.3 to 5.0]) and after multivariable adjustment, ECAC was not associated with total number of exacerbations (68 vs 56 events per 100 person-years; IRR, 1.12 [95% CI, 0.92 to 1.35]; P = .27) but was associated with a higher incidence of severe exacerbations (21 vs 17 events per 100 person-years; IRR, 1.39 [95% CI, 1.08 to 1.78]; P = .01).

Among participants without COPD (n = 4068; median follow up, 4.2 years [IQR, 3.0 to 4.9]) and after multivariable adjustment, ECAC was associated with increased frequency of both total number of exacerbations (54 vs 35 events per 100 person-years; IRR, 2.19 [95% CI, 1.78 to 2.71]; P < .001) and severe exacerbations requiring hospitalization (16 vs 10 events per 100 person-years; IRR, 2.95 [95% CI, 2.20 to 3.95]; P < .001).²¹

Among participants with COPD, after multivariable adjustment, there was no statistically significant association between the presence of ECAC and the time to first severe exacerbation (adjusted hazard ratio [HR], 1.28 [95% CI, 1.00 to 1.64]; P = .05) or the time to first any exacerbation (adjusted HR, 1.10 [95% CI, 0.94 to 1.28]; P = .24). Among participants without COPD, ECAC was associated with a shorter time to first severe exacerbation (adjusted HR, 1.35 [95% CI, 1.01 to 1.81]; P = .046) (eFigure 3 in the Supplement) but with no difference in time to first any exacerbation (adjusted HR, 1.09 [95% CI, 0.90 to 1.32]; P = .38).

Mortality data were available for 7389 participants, of whom 707 (9.6%) died during follow-up. Of those with ECAC, 38 (9.9%) died, compared with 669 (9.6%) of those without ECAC. After multivariable adjustment, ECAC was not associated with mortality (adjusted HR, 1.09 [95% CI, 0.77 to 1.54]; P = .64).

Expiratory scans were acquired at residual volume at a single center; the results for these participants are presented separately in eTables 1 and 2 in the Supplement. Overall, the prevalence of ECAC was higher in these participants compared with those who had functional residual capacity scans (9.4% vs 4.3%; P < .001; absolute difference, 5.1% [95% CI, 3.4% to 6.9%]). However, the relationships between respiratory morbidity and ECAC were similar, regardless of the level of expiration at which the scans were acquired (eTables 1 and 2 in the Supplement).

In this study of CT scans from a cohort of high-risk current and former smokers, the prevalence of ECAC was 5%, and presence of ECAC was associated with worse respiratory quality of life. Although previous smaller studies have ascribed respiratory symptoms to the presence of ECAC,^3,6,7 in this study ECAC was associated with respiratory morbidity independent of the degree of airflow obstruction and emphysema.

The prevalence of ECAC has not been extensively investigated, and previous studies were limited either by selection bias of patients with known respiratory problems who underwent bronchoscopy^2-4,22 or by small sample sizes.^4,23-26 ECAC has long been recognized to be associated with respiratory symptoms including chronic cough, dyspnea, stridor, and wheezing resistant to corticosteroids; impaired clearance of secretions; and respiratory failure.^7,27-29 However, these symptoms are nonspecific and have been variously ascribed to the underlying diseases such as chronic bronchitis and emphysema.³ We found that after adjustment for demographics, structural lung disease, and FEV₁, ECAC was associated with greater dyspnea and worse respiratory quality of life. We speculate that ECAC might explain some cases of dyspnea disproportionate to apparent obstructive airways disease measured by CT, spirometry, or both.⁸

We found that ECAC was associated with increased incidence of acute respiratory events on follow-up in participants without COPD and with increased incidence of severe exacerbations in those with airflow obstruction. This is a new finding and merits further investigation to determine whether ECAC represents a continuum of peripheral airway inflammation along the central airways or is an independent cause of respiratory decompensation. Although ECAC may be a marker of poorer outcomes in patients with COPD, that ECAC is associated with increased frequency of mild-moderate and severe respiratory events in those without airflow obstruction is also noteworthy. Recent large population studies have reported “exacerbation-like” events in participants without airflow obstruction.^21,30 Whether some of these represent decompensated ECAC or whether ECAC is a marker for future respiratory events needs to be investigated. Our results suggest that ECAC might contribute to symptoms independent of underlying disease and also may serve as a CT-based biomarker of poor respiratory outcomes.

ECAC in association with COPD has previously been attributed to a combination of the proximal extension of the inflammatory process of the peripheral airways,² weakening of membranous trachea from inflammation and cough,³¹ wall instability due to loss of parenchymal tissue,⁴ and increased pressure changes during exhalation.¹ We found no association between ECAC and any of the CT measures of COPD, suggesting that ECAC can occur without coexisting emphysema. However, the frequency of ECAC increased with COPD GOLD stage, suggesting an association with worsening lung disease, likely not discernible with existing CT protocols. We found a higher prevalence of chronic bronchitis among participants with ECAC than among controls.

Alternatively, this finding could reflect a lower FEV₁ associated with ECAC. Airflow obstruction in smokers has traditionally been thought to occur in the distal smaller airways less than 2 mm in diameter.⁵ However, it is plausible that severe central airway collapse is associated with additional airflow obstruction.³² There was no association between ECAC and gas trapping, a mechanism that could potentially lead to elevated intrathoracic pressures and collapse of weakened airway walls. That gas trapping does not have a major role in central airway collapse is supported by Lee et al,³³ who reported improvement in symptoms with tracheoplasty independent of the degree of air trapping. Previous reviews have suggested a link between cigarette smoking and ECAC.^7,22 Although smoking burden was significantly associated with ECAC in bivariable analyses, this relationship did not persist after consideration of underlying smoking-related disease. Greater BMI was associated with a greater frequency of ECAC. Obesity has been shown to affect the degree of tracheal collapse in patients with COPD,⁸ likely resulting from added pressure changes, similar to the mechanisms of increased small airway disease in obesity.³⁴

Our study has limitations. First, we used paired inspiratory-expiratory scans that were not obtained during dynamic expiration. Dynamic maneuvers cause greater collapse of the tracheobronchial walls.^35-37 However, we used a stringent 50% cutoff for CSA reduction despite scan acquisition with tidal exhalation and thus detected the more severe cases. Recent studies have shown that even volunteers free of respiratory disease can have significant airway collapse with dynamic scans,^36,37 and our measurements at tidal exhalation might be clinically more meaningful. Second, the COPDGene study includes smokers, with an oversampling of participants with COPD. Therefore, these findings may not be generalizable to the general population.

Third, many participants with COPD were receiving medications, and although the effect of medications on respiratory outcomes is modest, this may have influenced our results. We did not control for medication use, as only baseline information was known and we could not control for changing prescriptions, a particular issue for our assessment of exacerbations over time. We did observe that ECAC was associated with respiratory morbidity even in patients without COPD, who had lower rates of use of rescue and controller medications. Fourth, longitudinal data were available for only 7456 of the 8820 participants (413 of 443 with ECAC; 7043 of 8377 without ECAC).

Fifth, we analyzed the intrathoracic trachea alone and might have missed cases of upper airway malacia. Sixth, each scan was read by only a single reader, with subsequent calculation of interobserver agreement, which was fair to good. Seventh, we did not adjust for the change in lung volume with exhalation. However, this is difficult to control, because it is affected by both patient effort and underlying patient characteristics such as air trapping and lung elasticity. Spirometric gating was not used during the CT examinations. However, since ECAC is frequently underdiagnosed and is associated with respiratory symptoms, paired inspiratory-expiratory scans offer a simple method to diagnose ECAC and screen for severe cases.

In a cross-sectional analysis of current and former smokers, the presence of ECAC was associated with worse respiratory quality of life. Further studies are needed to assess long-term associations with clinical outcomes.

Corresponding Author: Surya P. Bhatt, MD, Division of Pulmonary, Allergy and Critical Care Medicine, University of Alabama at Birmingham, THT 422, 1720 Second Ave S, Birmingham, AL 35294 (spbhatt@uab.edu).

Author Contributions: Dr Bhatt had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Bhatt, Wise, Foreman, Dransfield.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Bhatt, Dransfield.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Bhatt, Curran-Everett, Dransfield.

Study supervision: All authors.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Bhatt reported receiving funding from the American Heart Association and NIH KL2 Scholarship, 1KL2TR001419. Dr Tschirren is an employee and shareholder at VIDA Diagnostics. Dr Lynch reported receiving consulting fees from Paraxel, Boehringer Ingelheim, Genentech, Gilead, and Intermun and grants from Siemens and the National Heart, Lung, and Blood Institute (NHLBI). Dr Wise reported receiving grant support and personal fees from Boehringer Ingelheim and GlaxoSmithKline; personal fees from Bristol-Myers-Squibb, Mylan, Novartis, Pfizer, Sunovion, Pulmonx, Spiration, Intermune, Grifols, and AstraZeneca; and grant support from Pearl Therapeutics and Forest Laboratories. Dr Washko reported holding consulting agreements with GlaxoSmithKline, Genentech, and Merck and that his spouse is an employee of Biogen. Dr Hoffman reported receiving grants from the National Institutes of Health (NIH) and that he is founder and shareholder in VIDA Diagnostics. Dr Dransfield reported receiving grants from the NIH, Department of Defence, and the American Heart Association; personal fees and other finding from Boehringer Ingelheim, Boston Scientific, and GlaxoSmithKline; and other funding from Pearl, Pulmonx, PneumRx, AstraZeneca, Novartis, and Yungjin. No other authors reported disclosures.

Funding/Support: This study was supported by awards R01HL089897, R01HL089856, and HL122438 from the NHLBI. The COPDGene project is also supported by the COPD Foundation through contributions made to an industry advisory board comprising AstraZeneca, Boehringer Ingelheim, Novartis, Pfizer, Siemens, Sunovion, and GlaxoSmithKline.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.